Add `hf` CLI model-discovery equivalent to generated Inference Providers task pages by davanstrien · Pull Request #2636 · huggingface/hub-docs

davanstrien · 2026-07-09T09:54:58Z

What

Following up on huggingface/huggingface_hub#4511 and this morning's Slack discussion — cc @Wauplin, opening as a draft for discussion since you gave it a 👍 ("it would make things easier for agents").

One-line change to TIP_LIST_MODELS_LINK_TEMPLATE in scripts/inference-providers/scripts/generate.ts + regeneration of the 20 task pages. Each page's "Explore all available models…" line under Recommended models now also shows the terminal equivalent:

hf models ls --warm --pipeline-tag automatic-speech-recognition --sort trending_score

--pipeline-tag (from #4512, thanks!) makes this an exact CLI mirror of the existing website filter link (models?inference=warm&pipeline_tag=<task>&sort=trending).

Draft until

--pipeline-tag ships in a huggingface_hub release — same gate as #2632. Will flip both to ready then.

Verification

Command tested against git-main hf for several tasks (returns the same warm model set as the website filter).
pnpm generate + pnpm format:check pass. Regeneration picked up routine data drift (token-classification output-spec table, zai-org GLM-5.2-FP8 mapping) and prettier fixed two pre-existing style nits in generate.ts — happy to drop any of that if you'd rather the bot take it.

Open questions

Wording/placement fine? Kept it as one continued sentence + code block rather than a separate TIP callout.
Keep --sort trending_score in the example, or shorter with just --warm --pipeline-tag <task>?

🤖 Generated with Claude Code

Note

Low Risk
Documentation and generator-template only; no runtime or API behavior changes beyond updated example model IDs in generated snippets.

Overview
Extends the generated Recommended models section on Inference Providers task docs so the existing Hub filter link is paired with a terminal equivalent: hf models ls --warm --pipeline-tag <task> --sort trending_score, linked to the hf CLI docs.

The change is driven by updating TIP_LIST_MODELS_LINK_TEMPLATE in scripts/inference-providers/scripts/generate.ts, then re-running the generator so all task markdown (and related provider pages) refresh.

Regeneration also reflects current warm-model/snippet data (e.g. HF Inference examples for image classification and text-to-image, zai-org VLM snippets using GLM-4.6V, and trimmed recommended-model bullets on several tasks). Minor formatting fixes in generate.ts are included.

^{Reviewed by Cursor Bugbot for commit 76c620b. Bugbot is set up for automated code reviews on this repo. Configure here.}

davanstrien · 2026-07-09T09:56:14Z

 <InferenceSnippet
    pipeline=text-generation
-    providersMapping={ {"zai-org":{"modelId":"zai-org/GLM-5.2","providerModelId":"glm-5.2"} } }
+    providersMapping={ {"zai-org":{"modelId":"zai-org/GLM-5.2-FP8","providerModelId":"glm-5.2"} } }


why u change this Claude

Not from the template edit — regeneration rewrites these pages from the live partners API, and zai-org's catalogue changed upstream: https://huggingface.co/api/partners/zai-org/models now lists zai-org/GLM-5.2-FP8 (isModelAuthor: true, listed first) alongside zai-org/GLM-5.2 for the same glm-5.2 endpoint, so the generator now picks the FP8 repo as the display model. Running pnpm generate on clean main reproduces this with zero local changes — the scheduled bot's next run would land the same. Kept it in (disclosed in the description), but happy to drop it and let the bot take it.

yes, not a problem

davanstrien · 2026-07-09T09:56:46Z

 <InferenceSnippet
    pipeline=text-generation
-    providersMapping={ {"cerebras":{"modelId":"openai/gpt-oss-120b","providerModelId":"gpt-oss-120b"},"cohere":{"modelId":"CohereLabs/tiny-aya-global","providerModelId":"tiny-aya-global"},"deepinfra":{"modelId":"zai-org/GLM-5.2","providerModelId":"zai-org/GLM-5.2"},"featherless-ai":{"modelId":"zai-org/GLM-5.2","providerModelId":"zai-org/GLM-5.2"},"fireworks-ai":{"modelId":"deepseek-ai/DeepSeek-V4-Pro","providerModelId":"accounts/fireworks/models/deepseek-v4-pro"},"groq":{"modelId":"openai/gpt-oss-120b","providerModelId":"openai/gpt-oss-120b"},"novita":{"modelId":"zai-org/GLM-5.2","providerModelId":"zai-org/glm-5.2"},"nscale":{"modelId":"meta-llama/Llama-3.1-8B-Instruct","providerModelId":"meta-llama/Llama-3.1-8B-Instruct"},"ovhcloud":{"modelId":"openai/gpt-oss-120b","providerModelId":"gpt-oss-120b"},"publicai":{"modelId":"swiss-ai/Apertus-8B-Instruct-2509","providerModelId":"swiss-ai/apertus-8b-instruct"},"scaleway":{"modelId":"meta-llama/Llama-3.1-8B-Instruct","providerModelId":"llama-3.1-8b-instruct"},"together":{"modelId":"deepseek-ai/DeepSeek-V4-Pro","providerModelId":"deepseek-ai/DeepSeek-V4-Pro"},"zai-org":{"modelId":"zai-org/GLM-5.2","providerModelId":"glm-5.2"}} }
+    providersMapping={ {"cerebras":{"modelId":"openai/gpt-oss-120b","providerModelId":"gpt-oss-120b"},"cohere":{"modelId":"CohereLabs/tiny-aya-global","providerModelId":"tiny-aya-global"},"deepinfra":{"modelId":"zai-org/GLM-5.2","providerModelId":"zai-org/GLM-5.2"},"featherless-ai":{"modelId":"zai-org/GLM-5.2","providerModelId":"zai-org/GLM-5.2"},"fireworks-ai":{"modelId":"deepseek-ai/DeepSeek-V4-Pro","providerModelId":"accounts/fireworks/models/deepseek-v4-pro"},"groq":{"modelId":"openai/gpt-oss-120b","providerModelId":"openai/gpt-oss-120b"},"novita":{"modelId":"zai-org/GLM-5.2","providerModelId":"zai-org/glm-5.2"},"nscale":{"modelId":"meta-llama/Llama-3.1-8B-Instruct","providerModelId":"meta-llama/Llama-3.1-8B-Instruct"},"ovhcloud":{"modelId":"openai/gpt-oss-120b","providerModelId":"gpt-oss-120b"},"publicai":{"modelId":"swiss-ai/Apertus-8B-Instruct-2509","providerModelId":"swiss-ai/apertus-8b-instruct"},"scaleway":{"modelId":"meta-llama/Llama-3.1-8B-Instruct","providerModelId":"llama-3.1-8b-instruct"},"together":{"modelId":"deepseek-ai/DeepSeek-V4-Pro","providerModelId":"deepseek-ai/DeepSeek-V4-Pro"},"zai-org":{"modelId":"zai-org/GLM-5.2-FP8","providerModelId":"glm-5.2"}} }


why this change Claude?

Same single upstream change as the zai-org provider page: this snippet's providersMapping is regenerated from the live partners API, where zai-org now advertises zai-org/GLM-5.2-FP8 for its glm-5.2 endpoint. Comes with any regeneration, not from the template edit — same story for the token-classification output-spec table rows.

HuggingFaceDocBuilderDev · 2026-07-09T09:58:53Z

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

davanstrien · 2026-07-09T10:00:31Z

also cc @hanouticelina for viz. Also fine if you prefer not to add this/do it differently.

One-line change to TIP_LIST_MODELS_LINK_TEMPLATE in generate.ts + regeneration: each task page's 'Explore all available models' line now also shows 'hf models ls --warm --pipeline-tag <task> --sort trending_score'. Regeneration picks up routine data drift (token-classification output spec, zai-org GLM-5.2-FP8 mapping) and prettier fixes in generate.ts. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Wauplin

Perfect!

And yeah, no worries about the re-generation, it's a process happening automatically every night. I'm merging it now to avoid further merge conflicts.

davanstrien commented Jul 9, 2026

View reviewed changes

davanstrien and others added 2 commits July 14, 2026 11:55

Regenerate task pages against current model data

76c620b

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

davanstrien force-pushed the ip-task-pages-cli-discovery branch from 3766406 to 76c620b Compare July 14, 2026 11:14

davanstrien marked this pull request as ready for review July 14, 2026 11:14

davanstrien requested review from Wauplin, hanouticelina and julien-c July 14, 2026 11:17

Wauplin approved these changes Jul 20, 2026

View reviewed changes

Wauplin merged commit 892dc27 into huggingface:main Jul 20, 2026
2 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Add `hf` CLI model-discovery equivalent to generated Inference Providers task pages#2636

Add `hf` CLI model-discovery equivalent to generated Inference Providers task pages#2636
Wauplin merged 2 commits into
huggingface:mainfrom
davanstrien:ip-task-pages-cli-discovery

davanstrien commented Jul 9, 2026 •

edited by cursor Bot

Loading

Uh oh!

davanstrien Jul 9, 2026

Uh oh!

davanstrien Jul 9, 2026

Uh oh!

Wauplin Jul 20, 2026

Uh oh!

davanstrien Jul 9, 2026

Uh oh!

davanstrien Jul 9, 2026

Uh oh!

HuggingFaceDocBuilderDev commented Jul 9, 2026

Uh oh!

davanstrien commented Jul 9, 2026

Uh oh!

Wauplin left a comment •

edited

Loading

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Uh oh!

Conversation

davanstrien commented Jul 9, 2026 • edited by cursor Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What

Draft until

Verification

Open questions

Uh oh!

davanstrien Jul 9, 2026

Choose a reason for hiding this comment

Uh oh!

davanstrien Jul 9, 2026

Choose a reason for hiding this comment

Uh oh!

Wauplin Jul 20, 2026

Choose a reason for hiding this comment

Uh oh!

davanstrien Jul 9, 2026

Choose a reason for hiding this comment

Uh oh!

davanstrien Jul 9, 2026

Choose a reason for hiding this comment

Uh oh!

HuggingFaceDocBuilderDev commented Jul 9, 2026

Uh oh!

davanstrien commented Jul 9, 2026

Uh oh!

Wauplin left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

davanstrien commented Jul 9, 2026 •

edited by cursor Bot

Loading

Wauplin left a comment •

edited

Loading